Understanding the approach
No. The protocol defines participant actions, information system rules and the limits of AI assistance. An instruction to a model alone does not create confirmed commitments.
A specific action: acknowledgment, agreement, commitment, authorisation or outcome acceptance. Each confirmation has content, a scope of responsibility and a conditions version.
Through shared rules in use, compatible records and tools, retained versions and delivery of changes to affected parties. The protocol text alone creates no network.
AI can support management without a common commitment protocol. ObliNet proposes explicit confirmations, dependencies, authority and revision across participants. Other systems can implement the same functions; the approach’s added value must be established by comparison.
ObliNet is a project for coordination through mutual commitments: who delivers a result to whom, by when, under which conditions and who accepts it. The links show dependencies on other participants’ actions.
Detailed answer →It specifies the expected result, parties, conditions and acceptance, and arises after participants clarify and confirm it. A task describes work; recording a commitment does not, by itself, make a promise a legal contract.
Detailed answer →Authority, resources, schedules and dependencies are checked before confirmation. When conditions change, affected participants agree further actions; the system must not rewrite commitments unnoticed on their behalf.
Detailed answer →It helps clarify conditions, find gaps and contradictions, and compare options. Its findings require checking. Consent, accepting commitments and confirming results remain within the authority of the relevant participants.
Detailed answer →The expected result, acceptance criteria, schedules, authority, resources, dependencies and confirmations from the parties are needed. If no change notification is received, the system may not learn of it; ready-made integrations and freedom from manual input are not promised.
Detailed answer →A hypothesis, draft models and protocols, and individual browser demonstrations have been prepared. The effectiveness of the complete system remains to be tested, accounting for benefits and all coordination costs.
Detailed answer →You can propose a recurring process, a situation or help with testing the hypothesis. Use the existing participation form. Sending an enquiry does not start a pilot or guarantee a response time.
Detailed answer →Understanding does not mean agreement. Agreement does not mean commitments are feasible. Fulfilling individual commitments does not guarantee achievement of the shared goal.
The proposed approach seeks a solution along several lines.
What exactly does each participant need? Do the parties understand one another’s requirements? Is mutual understanding and agreement on collaboration conditions possible in principle?
ObliNet seeks conditions for joint work in which each party obtains a result that matters to them while accepting agreed commitments.
“I need payment in advance” and “I will pay only after the result” look incompatible. But the first demand may reflect material costs and non-payment risk, while the second reflects the risk of losing money. These are hypotheses to clarify with the participants.
What is essential to each party? What can be conceded? Which additional conditions would make a concession acceptable? A participant may accept a lower price in exchange for guaranteed volume, or a later deadline in exchange for receiving results in stages.
In the payment example, these might include paying separately for materials, a small first stage, checking its result and then further funding. This changes the structure of collaboration itself: one party’s commitment becomes a condition for the other’s next action.
Who does what, and when? Are resources sufficient? How is the result accepted? What happens if there is a delay or circumstances change? Verbal acceptability does not mean the agreement will survive implementation.
If the initial balance of benefits, costs and risks changes, ObliNet should help reveal who is affected and which related commitments need renegotiating.
AI’s contribution — helping ask clarifying questions, compare conditions, propose options and detect contradictions. Participants themselves determine whether conditions are acceptable.
An acceptable agreement does not always exist. Resources may be insufficient, interests fundamentally incompatible, or the stronger party may benefit from imposing terms. Recording commitments does not, by itself, remove power imbalances or create trust.
A useful outcome is discovering in time that collaboration is not viable.
ObliNet is a research and engineering project working towards practical application. The next step is to test the proposed hypothesis experimentally, assess the approach’s benefits and identify conditions for its successful use. Research results will provide the basis for creating a platform to help people and organisations collaborate better.
Yes, you can view the example and do the exercises. Joint commitments require other parties to participate. A ready procedure for working with participants who are not connected has not yet been defined.
A joint process requires initial information and confirmations. AI assistance is intended to reduce routine work, but ready integrations and freedom from manual entry are not promised.
From a participant’s message or an agreed data source. If information does not arrive, the system may not learn of the change. In the educational schedule-change example, a button introduces the delay.
A record will not force them to do the work. It helps clarify expectations and preserve history; further action depends on agreements and existing procedures.
The participants, or an authorized mediator they choose, under a procedure agreed in advance. AI does not automatically become a judge. No ready dispute-resolution service is offered here.
The published privacy rules apply to the form. For the future platform and each pilot, access, storage and deletion must still be defined before data is transferred.
Approved pilot pricing and payment terms have not been published. A pilot should not be assumed to be free or paid by default; this must be agreed.
Possibly, but these pages do not present a confirmed outcome from a real deployment. Comparison must include both the benefit and all additional participant time.
You will send an enquiry to the project. Allow contact if you want a reply. Response time, a demonstration and participation in a pilot are not guaranteed in advance.
You can agree in a chat without ObliNet. The proposed addition is one current record of conditions and links to other agreements. It is useful only if it reduces searching, repetition and errors enough to justify the extra work.
Optional reading. This also describes proposed mechanisms and future capabilities; a description does not mean a feature is available. See “What is available now” for the current status.
What is available nowWhat ObliNet proposes, the role of AI, and what can be tested in practice.
ObliNet is a proposed coordination layer that makes explicit who commits to deliver what to whom, by when, under which conditions, and how the result will be confirmed. Links between commitments reveal what depends on other participants’ actions.
ObliNet organizes coordination around the commitment as the primary object of the system. Before acceptance, feasibility is checked: authority, resources, conditions and dependencies. After confirmation, execution, changing conditions and confirmation against agreed criteria are tracked. AI assists participants in this work; authority remains with the respective participants.
A task describes work, while a promise expresses an intention to do it. A confirmed commitment also specifies who may expect which result, under what conditions, and who accepts it. In ObliNet’s approach, it follows clarification and confirmation by the respective participants. Recording a promise does not by itself turn it into a legal contract.
Are the result and acceptance criteria clear? Does the participant have the authority, resources and time? Which permissions and actions by others are needed? Are those dependencies confirmed? Unknowns remain explicit: checking helps identify risk but does not guarantee success.
Yes. At the proposal stage, a participant can decline, request clarification or propose different terms. Identified infeasibility is a reason to revise the proposal before confirmation. Changing an already accepted commitment requires taking existing agreements and the parties’ authority into account.
Identify the affected commitment and those who depend on it, then agree on the next steps. Original terms, the reason for change and the confirmed revised agreement remain in the history. The system must not silently rewrite participants’ commitments.
Result criteria and acceptance authority are defined in advance. AI can compare evidence against criteria, but confirmation remains with the authorized party. In a disagreement, positions and evidence are recorded and the issue follows an agreed escalation process. A record marked “completed” is not proof of completion by itself.
It helps identify possible commitments in conversations, formulate conditions, find gaps and dependencies, compare options and assess evidence of fulfilment. Its conclusions can be wrong and must be open to verification. An AI assumption does not create the parties’ consent.
Technical capability alone does not confer authority. An agent may act on a participant’s behalf only within an explicit mandate. Beyond that mandate, a decision by the appropriate authorized participant is required. A model does not automatically gain the right to enter into or confirm commitments, or accept fulfilment.
No. Making commitments observable does not require exposing all correspondence or private life. Access should depend on role, authority and the purpose of verification. Each pilot must specify what information is collected, who can access it and how long it is retained.
Yes, that is a risk. Formalization is justified where the cost of ambiguity exceeds the effort of clarification. Start with one important interaction and a minimal set of fields. A pilot must measure both benefits and participants’ time, rather than treating more records as success.
In the proposed approach, an assignment is clarified into a confirmed commitment: who is responsible for the result, what they need from colleagues, when the work is ready and who accepts it. A manager can identify overload and incompatible deadlines earlier, while a participant can state conditions or propose a revision. For example, a designer confirms a deadline after agreeing when materials will arrive. Fewer missed commitments and less manual supervision are expected benefits to test, not established outcomes.
Commitments connect departments and independent participants: sales, production, procurement, logistics and the customer. A promised delivery date is checked against capacity, supplies and acceptance. A changed dependency reveals which agreements need revision. This complements existing accounting and management systems; authority is defined by the organisation’s charter, contracts and rules, rather than by AI.
A resident’s request can be connected to the responsible body’s authority, a specific commitment, funding, implementers and result criteria. Repairing a shelter, for example, depends on access, inspection, funding, work and acceptance. City officials see obstacles between services, while residents see the agreed result and available evidence of its achievement. Resident feedback complements formal acceptance within an established procedure. Closing a request should not substitute for solving the problem.
A public programme is represented as a network of linked commitments by ministries, regions, municipalities and implementers. Before a public promise, authority, resources, conditions and dependencies are clarified; afterward, execution and confirmation of results are tracked. AI helps detect conflicting plans and assess evidence. Political priorities, budget allocation and conflicts of interest remain matters for legitimate human and institutional decisions.
A commitment at one level becomes an explicit dependency at another. If a municipality relies on national funding, the funding conditions, responsible parties and confirmed status are recorded. A delay reveals which local deadlines and results are affected. Authorized participants agree on changes; the system must not automatically shift responsibility onto someone who does not control the cause of the delay.
The approach does not in itself entail such a replacement. It proposes making the connection between authority, commitments, actions and results observable within existing institutions and between independent participants. AI assists analysis but gains no right to appoint, punish or remove leaders on its own. Better accountability depends on reliable data, the right to challenge decisions, privacy protection and participants’ willingness to follow the rules.
Leaders should know that significant decisions, their grounds and consequences are available for assessment by those who delegated authority and by authorized oversight bodies. Delegating authority entails the ability to scrutinize its use in a company, institution, municipality or state. Disclosure is governed by reviewers’ authority and data protection requirements. A record does not prove that a decision was justified: evidence must be checked and assessments must be open to challenge. In public governance, this principle complements existing electoral and oversight institutions.
In business handoffs, client–contractor relationships, research projects, AI-agent workflows and the execution of public decisions. Repairing a facility, for example, depends on access, permits, funding, contractor work and acceptance. These are proposed applications, not claims of proven effectiveness across these settings.
The scientific hypothesis, draft models and protocols for comparative experiments, and individual browser demonstrations have been prepared. The next step is to test the proposed hypothesis experimentally, assess the approach’s benefits and identify conditions for its successful use. The research results will provide the basis for creating a platform. A working demonstration does not establish the effectiveness of the future platform as a whole.
Choose one recurring process, its participants and a problem, such as a missed dependency or ambiguous acceptance. Record baseline measures and evaluation criteria. Compare explicit commitments with the previous process: timing, misunderstandings, renegotiations, confirmed results and time spent.
Use the Get involved form: enter your email, select your interests and write a message. No registration is needed. If you want a reply, give permission to contact you. Consent to receive news is a separate choice.